Built an MCP tool called Ownhand. Your Agent can hook into it automatically when writing, it writes content with the lowest slop score of any model.

A developer built Ownhand, an MCP tool that lets AI agents hook in automatically when writing and adapt to a user's style by learning implicitly from edited versions of messages. The tool accepts the domain, recent message history, extra context and recipient as call parameters, and the developer claims it produces content with the lowest slop score of any model. It is available to try without an account at ownhand.dev.

This idea was born during the Opus 5 era please, never again , but it's still useful today I use it dozens of times a day as a freelance consultant . It's entirely AI native, so you don't have to do anything, your agent handles it. It grows with you as you use it, and you can even create custom domains, i.e. reddit comments, texts to family members jk . You teach it by simply sending it the edited version what you sent irl and it learns implicitly from that. Your agent makes calls to the MCP with the domain, recent message history, extra context, and even who it's being sent to. Ownhand handles all of it. I'm wondering how you guys handle this issue; I'd love to hear your approaches. I've tried skills before, but it's just really bad at it, even with Opus 5.5. Try it account free at https://ownhand.dev/try https://ownhand.dev/try